He et al. World J Surg Onc (2021) 19:224 https://doi.org/10.1186/s12957-021-02342-y

RESEARCH Open Access Pan‑cancer analysis of ­m5C regulator reveals consistent epigenetic landscape changes in multiple cancers Yuting He1,2,3,4* , Xiao Yu1,2,3,4, Menggang Zhang1,2,3,4 and Wenzhi Guo1,2,3,4*

Abstract Background: 5-Methylcytosine ­(m5C) is a reversible modifcation to both DNA and various cellular RNAs. However, its roles in developing human cancers are poorly understood, including the efects of mutant m­ 5C regulators and the outcomes of modifed nucleobases in RNAs. Methods: Based on The Cancer Genome Atlas (TCGA) database, we uncovered that mutations and copy number variations (CNVs) of ­m5C regulatory genes were signifcantly correlated across many cancer types. We then assessed the correlation between the expression of individual ­m5C regulators and the activity of related hallmark pathways of cancers. Results: After validating ­m5C regulators’ expression based on their contributions to cancer development and pro- gression, we observed their upregulation within tumor-specifc processes. Notably, our research connected aberrant alterations to ­m5C regulatory genes with poor clinical outcomes among various tumors that may drive cancer patho- genesis and/or survival. Conclusion: Our results ofered strong evidence and clinical implications for the involvement of m­ 5C regulators. Keywords: m5C regulatory genes, Frequent network mining, Pan-cancer analysis, Survival, 5-Methylcytosine

Background are crucial factors and contribute to cancer pathogenesis Cancers have become the second life-threatening malig- [10–12]. Methylation is an essential epigenetic modifca- nancies, which contribute to almost 18.1 million people tion and is closely related to the pathogenesis of cancers occurred and 9.6 million death globally in 2018 [1]. Lack [13–17]. Te 5-methylcytosine ­(m5C), N6-methyladenine of efcient diagnosis indicators at an early stage and high ­(m6A), and N1-methyladenosine ­(m1A) have become rate of postoperative recurrence contribute to poor clini- the most common types of epigenetic modifcations in cal prognosis and high mortality [2, 3]. Growing evidence eukaryotes [13]. Emerging evidence has demonstrated demonstrated that genomic instability [4, 5], oncogene that ­m5C modifcation has the potential to serve as novel activation, aberrant methylation modifcations, altera- epigenetic markers with remarkable biological signif- tions in epigenetic changes [6–8], aberrant expression cance in biological processes [18–20]. of microRNAs [9], and alterations of signaling pathways m5C modifcation distributes in diferent types of RNAs and DNAs [21–23]. ­m5C modifcations can even modify the destiny of cancer cells [24]. ­m5C regulators *Correspondence: [email protected]; [email protected] 1 Department of Hepatobiliary and Pancreatic Surgery, The First Afliated contain writers, erasers, and readers, which function as Hospital of Zhengzhou University, No.1 Jianshe Road, Zhengzhou 450052, common epigenetic modifcation and contribute to pre- China mRNA splicing, expression, gene silencing, nuclear Full list of author information is available at the end of the article export, genomic maintenance, and translation initiation

© The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​ iveco​ mmons.​ org/​ licen​ ses/​ by/4.​ 0/​ . The Creative Commons Public Domain Dedication waiver (http://creat​ iveco​ ​ mmons.org/​ publi​ cdoma​ in/​ zero/1.​ 0/​ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data. He et al. World J Surg Onc (2021) 19:224 Page 2 of 12

modifcations [25, 26]. ­m5C could therefore be used as including somatic mutations, copy number variations a biomarker for disease progression, including various (CNVs), gene expression, and RNA-seq data (Fig. 1B). types of cancers [27]. ­m5C maintains open and closed 5 chromatin states to control gene expression, genome Genomic data collection of ­m C regulators editing, organismal development, and cellular diferen- Tirteen ­m5C regulators were identifed from published tiation [23]. In this context, writers act within a meth- papers. Information of ­m5C regulators was collected yltransferase complex to methylate targets, and erasers from Gene Cards (www.​genec​ards.​org). Ensemble gene remove ­m5C methylation, while readers recognize and IDs and HUGO Committee symbols bind to ­m5C-methylated RNA and implement corre- were assigned to each ­m5C regulator-associated gene. sponding functions [25, 28, 29]. Te anomalous interplay between writers and erasers, arising from alterations to Whole genomics data analysis of 33 pan‑cancers their expression, has been linked to cancer pathogenesis Te integrated OMICS datasets based on the TCGA and progression [23, 30]. However, pan-cancer efects database of 33 pan-cancers were applied in this study. of changes to m­ 5C regulatory gene expression have not We collected the mutation annotation format profle been fully defned. Next-generation sequencing (NGS) from TCGA, which contains over 10,000 cancer patient’s provides us efective tools to comprehensively view the information. Te level 3 data of copy number alterations ­m5C distribution landscape throughout the global tran- profles in TCGA was acquired for secondary analysis. scriptome [27]. In addition, we downloaded pan-cancers RNAseq data In this study, we identifed the potential prognostic of genomic variations profles and corresponding clini- value of m­ 5C regulators and provided a comprehensive cal information from the Genomic Data Commons Data understanding of ­m5C modifcations in pan-cancers, Portal using the R package “TCGAbiolinks.” which will help to fnd novel opportunities for cancer early detection, treatment, and prevention. Diferently expression genes (DEGs) identifcation Te DESeq package in R language was utilized to validate Materials and methods the DEGs between 33 pan-cancer samples and adjacent Study workfow samples. Genes with a mean value > 0 were included in We downloaded fragments of kilobase transcripts based the screening of DEGs. To establish proper DEGs among on fragments per kilobase of transcript per million 33 cancers, we settled the adjusted P value less than 0.01 (FPKM) gene expression from Te Cancer Genome Atlas and |log2 fold change (log2Fc) | no less than two as the (TCGA, https://​www.​cancer.​gov/) dataset among 33 dif- statistical threshold value for diferentially expressed ferent cancer types. ­m5C regulator patterns were inves- genes. Te results were screened as signifcant DEGs and tigated in 5480 samples among 33 diferent cancer types, methylated sites.

Fig. 1 m5C regulators and the function in cancers. A. The distribution and the function of m­ 5C regulatory writers, eraser, and reader. B The workfow scheme for this study He et al. World J Surg Onc (2021) 19:224 Page 3 of 12

Functional annotations and pathway enrichment analysis DNMT3A, and DNMT3B), one eraser (TET2), and one To evaluate the biological functions of each ­m5C modif- reader (ALYREF). Tis study validated the frequency of cation-related gene, we transformed the RNA-seq data of ­m5C regulator patterns among 33 cancers by integrating all samples into transcripts per million (TPM) values. Te somatic mutations and CNVs data. Table 1 illustrated methods have been described in a previous study [31]. Te detailed information. Te results indicated that the over- insufcient, duplicated, and zero expression genes will be all average mutation frequency of regulatory factors is eliminated. Furthermore, the gene set variation analysis low, ranging from 0 to 9% (Fig. 2A). Te uterine corpus (GSVA) was applied to determine transcriptomic activities endometrial carcinoma (UCEC) is characterized as a high and explore the biological processes of ­m5C regulators. To tumor mutation burden [33]. Te UCEC showed signif- further explore m­ 5C regulators related inhibition and acti- cantly higher mutation frequency. Horizontal analysis vation factors, we performed the Pearson correlation coef- indicated that TET2, DNMT3B, DNMT3A, and DNMT1 fcient (PCC) and defned the absolute value of the PCC demonstrated much higher mutation frequency among greater than 0.5 and p value of less than 0.01 as the screen 33 cancers. Furthermore, we uncovered the CNV muta- cut-of. Te results could be recognized as signifcantly cor- tion frequency of ­m5C regulators was common. Regula- related m­ 5C regulators. tors such as DNMT3B, ALYREF, and NSUN5 displayed extensive CNVs. On the contrary, TET2 and NSUN4 5 The internships between ­m C regulators showed signifcant lack of ­m5C modifcation related CNV To visualize the intercorrelations among m­ 5C regulators, mutations among pan-cancers (Fig. 2B, and Table 2). we adopted the “CORPRRAP” R package (https://github.​ ​ To further investigate whether the genomic mutations com/taiyun/​ corrp​ lot​ ). Besides, the STRING database was afect ­m5C regulators expression, we intensively detected also applied for the exploration and analysis of these associ- the ­m5C regulators’ gene expression disturbances in ations between ­m5C regulators and 325 related genes [32]. thirty-three pan-cancers and fve standard control sam- Te 325 genes were obtained from the Kyoto Encyclope- ples. Te result implied that the CNV alterations (ampli- dia of Genes and Genomes (KEGG) database (http://www.​ ​ fcation and deletion) might profoundly afect the ­m5C kegg.jp/​ or http://www.​ genome.​ jp/​ kegg/​ ). Te correlations regulator’s expression (Fig. 3A). Te m­ 5C regulators between m­ 5C regulators and 325 genes were visualized with CNV amplifcation showed signifcantly increased through Cytoscape (https://cytos​ cape.​ org/​ ). expression in pan-cancers (such as DNMT3B), and m­ 5C regulators with CNV deletion exhibited remarkably 5 Clinical characteristic of m­ C regulators decreased expression, like TET2. In addition, we com- To explore the m­ 5C regulators’ related clinical charac- paratively analyzed the m­ 5C regulators’ expression levels teristics, we classifed genes into high and low expression in cancers and corresponding normal tissues and found groups based on genes’ median expression. Correlations out that DNMT3B was signifcantly overexpressed in between outcomes of the two groups were then analyzed thirty-three tumor or cancer tissues compared with adja- through a log-rank test via R software (https://cran.r-​ proje​ ​ cent normal tissues (Fig. 3B). Tese results uncovered ct.org/​ web/​ packa​ ges/​ survi​ val/​ index.​ html​ ). Te log-rank that the ­m5C regulators among various cancers showed test was performed to weigh the overall survival rates signifcant heterogeneity in gene expression and genet- that difer between the high and low expression groups. ics. Collectively, our results demonstrated that aberrant Te CRAN Package survival (https://cran.r-​ proje​ ct.​ org/​ ​ ­m5C regulations were crucial for carcinogenesis and pro- web/packa​ ges/​ survi​ val/​ index.​ htm​ ) was performed, and gression, which provided a clue for further functional we defned the p value of less than 0.05 as signifcant detection. diference. m5C regulators related pan‑carcinogenic pathways The roles of ­m5C regulators in cell growth To comprehensively explore the molecular mechanism Te CRISPR-CAS9 gene scale screening of cell lines from ­m5C regulators involved in cancers, we evaluated the 33 cancer types was collected from previous study [32]. We correlations between ­m5C regulators’ expres- calculated the proportion of every regulator as an essential sion and KEGG enrichment analysis-related activities. gene in the cell lines. Te results indicated that m­ 5C regulator proteins have Results a close relationship with tumor-related pathways’ acti- vation and inactivation (Fig. 4A, and Table 3). ALYREF, Results m5C regulators identifcation and its genomic DNMT1, and TET2 were involved in the cell cycle, extensive genetic changes 5 DNA replication, and prostate cancer-related pathways. In this study, we identifed 13 ­m C regulators, as shown Notably, ALYREF was involved in multiple pathways, in Fig. 1A, eleven writers (NSUN1-7, DNMT1, DNMT2, including cell cycle, DNA replication, and prostate He et al. World J Surg Onc (2021) 19:224 Page 4 of 12

Table 1 The 33 cancer types in TCGA pan-cancer project Cancer types Abbr Normal tissues Cancer tissues Mutation CNV

Kidney Renal Clear Cell Carcinoma KIRC 72 539 370 531 Kidney Renal Papillary Cell Carcinoma KIRP 32 289 282 291 Kidney Chromophobe KICH 24 65 66 69 Brain Lower Grade Glioma LGG 0 529 526 516 Glioblastoma Multiforme GBM 5 169 403 580 Breast Invasive Carcinoma BRCA​ 113 1109 1026 1083 Lung Squamous Cell Carcinoma LUSC 49 502 485 504 Lung Adenocarcinoma LUAD 59 535 569 519 Rectum Adenocarcinoma READ 10 167 151 168 Colon Adenocarcinoma COAD 41 480 408 454 Uterine Carcinosarcoma UCS 0 56 57 59 Uterine Corpus Endometrial Carcinoma UCEC 35 552 531 542 Ovarian Serous Cystadenocarcinoma OV 0 379 412 582 Head and Neck Squamous Carcinoma HNSC 44 502 509 525 Thyroid Carcinoma THCA 58 510 500 502 Prostate Adenocarcinoma PRAD 52 499 498 495 Stomach Adenocarcinoma STAD 32 375 439 444 Skin Cutaneous Melanoma SKCM 1 471 468 370 Bladder Urothelial Carcinoma BLCA 19 414 411 411 Liver Hepatocellular Carcinoma LIHC 50 374 365 373 Cervical Squamous Cell Carcinoma and Endocervical CESC 3 306 291 298 Adenocarcinoma Adrenocortical Carcinoma ACC​ 0 79 92 93 Pheochromocytoma and Paraganglioma PCPG 3 183 184 165 Sarcoma SARC​ 2 263 239 260 Acute Myeloid Leukemia LAML 0 151 141 194 Pancreatic Adenocarcinoma PAAD 4 178 178 187 Esophageal Carcinoma ESCA 11 162 185 187 Testicular Germ Cell Tumors TGCT​ 0 156 151 153 Thymoma THYM 2 119 123 126 Mesothelioma MESO 0 86 82 90 Uveal Melanoma UVM 0 80 80 83 Lymphoid Neoplasm Difuse Large B-cell Lymphoma DLBC 0 48 37 51 Cholangiocarcinoma CHOL 9 36 36 39 In total 730 10,363 10,295 10,944

cancer-related pathways. DNMT1 was involved in drug with each other. Te eraser TET2 was signifcant corre- metabolism, lipid metabolism, and nucleic acid bio- lated with the writer NSUN3. Writers such as ALYREF synthesis signaling pathways [34]. ALYREF, DNMT1, and NSUN5 also showed obvious correlations (R = 0.55, NSUN5, NSUN1, and TET2 showed active involvement P < 0.01) (Fig. 4C). Additionally, to visualize the interac- in KEGG enrichment pathways (Fig. 4B). In addition, tions between m­ 5C regulators, we utilized the – genes will not function alone [35]. Growing evidence protein interaction (PPI) analysis in ­m5C regulators indicated that genes always co-efect with multiple related proteins. Te results showed that writers, read- genes and always have multiple functions [35, 36]. We ers, and erase were particularly frequent (Fig. 4D). further explored the internal connections between ­m5C Tese results indicated that interactions among ­m5C regulators gene expression. Results indicated that the regulators play crucial roles in the development and readers, writers, and erasers also have high correlations progression of cancers. He et al. World J Surg Onc (2021) 19:224 Page 5 of 12

Fig. 2 Mutation and CNV of ­m5C regulators across pan-cancer. A The mutant frequency of ­m5C regulators across 33 cancer types. B CNV analysis of ­m5C regulators across cancer types. The upper part of each grid shows the deletion frequency, and the bottom part shows the amplifcation frequency

Table 2 The mutation frequency of m5C regulators across 33 cancer types (Top 5) Function Genes UCEC SKCM COAD READ STAD writers NSUN1 0 0 0 0 0 NSUN2 0.0622642 0.0192719 0.0250627 0.0218978 0.020595 NSUN3 0.0283019 0.0021413 0.0100251 0.0145985 0.0091533 NSUN4 0.0188679 0.0085653 0.0125313 0 0.006865 NSUN5 0.0207547 0.0171306 0.0200501 0 0.006865 NSUN6 0.045283 0.0149893 0.0150376 0.0145985 0.0183066 NSUN7 0.0509434 0.0449679 0.0125313 0.0072993 0.0091533 DNMT1 0.0773585 0.0428266 0.0401003 0.0437956 0.0320366 DNMT2 0 0 0 0 0 DNMT3A 0.0660377 0.0278373 0.0225564 0.0218978 0.0183066 DNMT3B 0.0811321 0.0342612 0.037594 0.0510949 0.0343249 eraser TET2 0.0943396 0.0428266 0.0551378 0.0291971 0.0320366 reader ALYREF 0.0188679 0.0021413 0.0050125 0.0072993 0.0022883 He et al. World J Surg Onc (2021) 19:224 Page 6 of 12

Fig. 3 The association between CNV and the gene expression of ­m5C regulatory genes. A Alterations to ­m5C regulatory gene expression in 24 cancer types. The heat map demonstrates fold change, with red representing upregulated genes and blue representing downregulated genes. B Box plots exhibit the expression distribution of DNMT3B across tumor and normal samples in 24 cancer types

5 Clinical signifcance of ­m C regulators in pan cancers the prediction of PFI at fxed time points in patients with To evaluate the clinical prognosis of ­m5C regulators, thirty-two solid tumors. PFIs of DNMT3B and DNMT1 we calculated the overall survival (OS), overall median showed signifcantly higher hazard ratio values, indicat- progression-free interval (PFI), disease-specifc survival ing that they have the potential to be utilized as unfavora- (DSS), and disease-free interval (DFI) of ­m5C regulators. ble prognosis prediction factors. Te PFI in ACC, KICH, Te OS analysis implied a signifcant correlation between KIRC, LGG, and uveal melanoma (UVM) showed signif- ­m5C regulators and thirty-three pan-cancers. Te heat cant correlation ships with most ­m5C regulators (Fig. 5B). map demonstrated that ­m5C regulators were signifcantly Similarly, Te DSS and DFI analyses indicated that ACC correlated with survival of patients, including OS, PFI, and LGG showed remarkably correlations with most ­m5C DSS, and DFI. In detail, the OS in adrenocortical car- regulators (Fig. 5C, D). Tese results indicated that ­m5C cinoma (ACC), kidney chromophobe (KICH), kidney regulators have crucial prognostic prediction values in a renal clear cell carcinoma (KIRC), brain lower grade gli- variety of cancer types. oma (LGG), and liver hepatocellular carcinoma (LIHC) 5 showed signifcant correlations with ­m5C regulators Efect of m­ C regulators in LIHC and cholangiocarcinoma (Fig. 5A). Te highly expressed DNMT3A, DNMT3B, (CHOL) DNMT1, and ALYREF were signifcantly related to Studies have validated that ­m5C-related genes altera- poor prognosis. Collectively, the DNMT3A, DNMT3B, tions have a close relationship with advanced tumor DNMT1, and ALYREF might function as poor prognosis progression and advanced tumor stages [37], based on predictors in cancer progression. Moreover, we evaluated the above pieces of evidence that most m5C regulators He et al. World J Surg Onc (2021) 19:224 Page 7 of 12

Fig. 4 m5C regulators are associated with the activation and inhibition of cancer pathways. A Network landscape demonstrating the correlation between ­m5C regulators and cancer pathways. Red represents a positive correlation, and blue represents a negative correlation. The size of the nodes corresponds to the number of links. B The number of pathways correlated with individual m­ 5C regulators. The upper panel represents positively correlated pathways, and the bottom panel represents negatively correlated pathways. C The correlation among the expression of m­ 5C regulators. D The PPI network of ­m5C regulators

are associated with patients’ OS in LIHC and CHOL Discussion (Fig. 5A). Based on the overall expression patterns of Epigenetic variation is often related to human disease, ­m5C regulators, all patients in these cancer groups were especially cancers [12, 38–40]. Remarkably progression categorized into two subgroups. Te frst subgroup con- has been made of various epigenetic-targeted therapies sisted of 112 patients indicating high expression of ­m5C that have a broad application of malignancies and have regulators (Reg-high), and the second subgroup con- exhibited detection and therapeutic potential for solid sisted of 295 patients with low m­ 5C regulators expres- tumors in preclinical and clinical trials [41–45]. Aberrant sion (Reg-low) (Fig. 6A). Compared with the Reg-high methylation regulators process both in DNA and RNA subgroup, the survival probability of patients in the play a critical role in epigenetic regulators, which are sig- Reg-low subgroup was signifcantly better (P < 0.001) nifcantly associated with tumorigenesis [17, 46–51]. (Fig. 6B). Tese results indicated that ­m5C regulators Original reports described that NSUN2 participates have a potential function as prognostic indicators in in catalyzing biological reactions of ­m5C formation in hepatocellular carcinoma and cholangiocarcinoma. RNAs and regulating cell cycle [52], linked to stem cell He et al. World J Surg Onc (2021) 19:224 Page 8 of 12

Table 3 The CNV-Gain and CNV-loss frequency of m5C regulators across 33 cancer types (Top 5) Genes CNV Gain CNV loss KICH OV ACC​ UCS LUSC KICH ACC​ TGCT​ UCS OV

NSUN1 0 0 0 0 0 0 0 0 0 0 NSUN2 0.560606 0.558678 0.633333 0.464286 0.699801 0.227273 0.144444 0.544872 0.125 0.099174 NSUN3 0.545455 0.540496 0.122222 0.232143 0.526839 0.212121 0.411111 0.102564 0.142857 0.044628 NSUN4 0.015152 0.499174 0.033333 0.339286 0.073559 0.863636 0.6 0.128205 0.089286 0.087603 NSUN5 0.818182 0.477686 0.511111 0.25 0.26839 0.030303 0.022222 0.032051 0.178571 0.082645 NSUN6 0.075758 0.428099 0.255556 0.321429 0.089463 0.787879 0.3 0.416667 0.321429 0.135537 NSUN7 0.863636 0.195041 0.411111 0.25 0.083499 0.015152 0.111111 0.49359 0.285714 0.408264 DNMT1 0.727273 0.418182 0.577778 0.285714 0.101392 0.015152 0.033333 0.198718 0.375 0.295868 DNMT2 0 0 0 0 0 0 0 0 0 0 DNMT3A 0.045455 0.403306 0.111111 0.464286 0.335984 0.772727 0.422222 0.012821 0.017857 0.102479 DNMT3B 0.787879 0.689256 0.555556 0.660714 0.39165 0.045455 0.111111 0.038462 0.017857 0.019835 TET2 0.818182 0.044628 0.366667 0 0.037773 0.015152 0.122222 0.608974 0.642857 0.694215 ALYREF 0.030303 0.320661 0.166667 0.464286 0.252485 0.787879 0.411111 0.038462 0.125 0.292562

diferentiation and involved in progression [53]. It is also may signifcantly increase the risk of cancer [62]. In this reported that ­m5C regulators such as NSUN2 and bind- research, the DNMT3B, ALYREF, and NSUN5 showed ing partner ALYREF participant in promoting mRNA extensive CNV amplifcation. In contrast, CNVs such as export coordinately [54], and NSUN6, in complex with a TET2 and NSUN4 are generally deletion. Tese results full-length tRNA substrate targeting cytosine accessible indicated that CNV and the associated gene signatures to the enzyme for methylation [14, 55, 56]. Te NSUN3 is are useful for early cancer detection and diagnosis, tar- required for the deposition of ­m5C at the anticodon loop geted therapeutics, and prediction of prognosis. in the mitochondria encoded transfer RNA methionine Te genomic and transcriptomic parameters of various [57]. Te NSUN5 demonstrates suppression character- cancers are associated with ­m5C regulators gene expres- istics in vivo glioma models [58]. NSUN5 gene mutation sion and activity of the KEGG pathways [63]. We also leads to an un-methylated condition at the C3782 posi- investigate the gene expression perturbations of ­m5C reg- tion of 28S rRNA, which leads to a total depletion of ulators through 33 cancer types with parallel normal con- protein synthesis and inducing an adaptive translational trols. Te expression of ALYREF, DNMT1, NSUN2, and program under stress collectively [59], as is illustrated TET2 are more positively correlated with the majority of that the m­ 5C regulators may infuence a wide variety of pathways, such as the cell cycle, DNA replication, spli- biological functions and metabolism. ceosome, and nucleotide excision repair pathways. Te In the present study, we applied certain methodo- DNMT1 expression is related to the activation of mul- logical particularities to build a model and evaluated tiple metabolic pathways, including drug metabolism, a catalog of genomic characteristics of tumors associ- lipid metabolism, and nucleic acid biosynthesis signal- ated with m­ 5C regulators. We obtained a total of 13 ­m5C ing pathways. Te ­m5C regulators’ pathways are signif- regulators. Te mutations and CNVs of ­m5C regulators cantly essential for a wide range of biological processes. are linked to several tumor developments. All cancers m5C regulators were also validated involved in malignant carry somatic mutations [60]. UCEC exhibited a signif- activities [64]. Recent studies have demonstrated that the cantly higher number of mutations across pan-cancers, m5C modifcation in pyruvate kinase muscle isozyme M2 analogously TET2, DNMT3B, DNMT3A, and DNMT1 was involved in bladder cancer proliferation and migra- have placed a moderate burden in ­m5C regulator genes. tion. M5C regulator Aly/REF export factor regulated DNMT3B gene mutation was generally higher expression pyruvate kinase muscle isozyme M2 promote the glucose level among various cancers. Here, we sequenced the metabolism of bladder cancer [64]. At the same time, the ­m5C regulator genomes of pan-cancer and providing the precise molecular modifcation mechanisms and cellu- frst comprehensive remarkable insights into the forces lar processes among pan-cancer need further study and that have shaped various cancer genomes. CNVs play an deeper exploration for a better prognosis. important role in tumor genesis and progression [61], For a deeper exploration of the relationship between including amplifcation and deletion of oncogenes, which ­m5C regulators and their clinical outcomes, we describe He et al. World J Surg Onc (2021) 19:224 Page 9 of 12

Fig. 5 Summary of the relationship between ­m5C regulators expression and patient’s survival. A Overall survival (OS) of ­m5C regulators across 33 cancer types. B Progression-free interval (PFI) of ­m5C regulators across 32 solid cancer types. C Disease-specifc survival (DSS) of m­ 5C regulators across 32 solid cancer types. D Disease-free interval (DFI) of ­m5C regulators across 28 solid cancer types. Red represents a higher ­m5C regulator expression associated with poor survival, and blue represents an association with better survival a comprehensive landscape of m­ 5C regulator path- relationship between cancer-associated clinical relevance ways activities across diferent cancer types and identify and ­m5C regulators. To determine the efect of methyla- cancer characteristics in relation to clinical outcome. tion-based molecular for earlier detection diagnostics in Collectively, we provide robust evidence for the close patients with several types of cancer, we systematically He et al. World J Surg Onc (2021) 19:224 Page 10 of 12

Fig. 6 Efect of ­m5C regulators on patients with hepatocellular carcinoma and cholangiocarcinoma. A Heat map showing clustering for CHOL and LIHC patients based ­m5C regulator expression. Yellow represents Reg-low subgroup (N 295), and green represents Reg-high subgroup (N 112). B = = Kaplan–Meier survival plot of patients grouped by global ­m5C regulator expression pattern (P < 0.001)

analyzed the ­m5C regulators’ pathway activities with Authors’ contributions YH and WG defned the research theme and discussed analyses, interpreta- the functional and clinically complication for estimating tion, and presentation. XY and MZ drafted the manuscript and analyzed the tumor development and progression with potential prog- data. MZ helped with references collection. The authors read and approved nostic value. the fnal manuscript. Funding This work was supported by the National Natural Science Foundation of Conclusion China (81902832) and the Youth Talent Lifting Project of Henan Province (2021HYTP059). Te m5C regulators were diferently expressed and showed signifcantly diferent CNVs in pan-cancers, Availability of data and materials which also involved multiple oncogene pathways. In All of the data involved in this study are available in the public databases which are listed in the “Materials and methods” section. addition, m5C regulators also exhibited prognosis predic- tion value in pan-cancers. Terefore, our study provides Declarations a better understanding of the biology of m5C regulators in pan-cancers, indicating that m5C RNA methylation Ethics approval and consent to participate regulators have the potential to become novel biomarkers This was not applicable to this manuscript. and therapeutic targets for various tumors. Consent for publication Consent for publication was obtained from all participants.

Abbreviations Competing interests m5C: 5-Methylcytosine; TCGA​: The Cancer Genome Atlas; CNVs: Copy The authors declare that they have no competing interests. number variations; m6A: N6-methyladenine; m1A: N1-methyladenosine; NGS: Next-generation sequencing; FPKM: Fragments per kilobase of transcript Author details per million; DEGs: Diferently expression genes; TPM: Transcripts per million; 1 Department of Hepatobiliary and Pancreatic Surgery, The First Afliated Hos- GSVA: Gene set variation analysis; PCC: Pearson correlation coefcient; UCEC: pital of Zhengzhou University, No.1 Jianshe Road, Zhengzhou 450052, China. Uterine corpus endometrial carcinoma; KEGG: Kyoto Encyclopedia of Genes 2 Key Laboratory of Hepatobiliary and Pancreatic Surgery and Digestive Organ and Genomes pathway; PPI: Protein–protein interaction; OS: Overall survival; Transplantation of Henan Province, The First Afliated Hospital of Zhengzhou PFI: Progression-free interval; DSS: Disease-specifc survival; DFI: Disease-free University, Zhengzhou 450052, China. 3 Open and Key Laboratory of Hepato- interval; ACC:​ Adrenocortical carcinoma; KICH: Kidney chromophobe; KIRC: biliary & Pancreatic Surgery and Digestive Organ Transplantation At Henan Kidney renal clear cell carcinoma; LGG: Brain lower grade glioma; LIHC: Liver Universities, Zhengzhou 450052, China. 4 Henan Key Laboratory of Digestive hepatocellular carcinoma; HR: Hazard ratio; UVM: Uveal melanoma; CHOL: Organ Transplantation, Zhengzhou 450052, China. Cholangiocarcinoma. Received: 23 April 2021 Accepted: 21 July 2021 Supplementary Information The online version contains supplementary material available at https://​doi.​ org/​10.​1186/​s12957-​021-​02342-y. References 1. Cortes J, Perez-García JM, Llombart-Cussac A, Curigliano G, El Saghir NS, Additional fle 1. Cardoso F, Barrios CH, Wagle S, Roman J, Harbeck N, et al. Enhancing global access to cancer medicines. CA Cancer J Clin. 2020;70:105–24. 2. van der Pol Y, Mouliere F. Toward the early detection of cancer by decod- Acknowledgements ing the epigenetic and environmental fngerprints of cell-free DNA. Not applicable. Cancer Cell. 2019;36:350–68. He et al. World J Surg Onc (2021) 19:224 Page 11 of 12

3. Hu T, Wolfram J, Srivastava S. Extracellular vesicles in cancer detection: 29. Reid R, Greene PJ, Santi DV. Exposition of a family of RNA m(5)C methyl- hopes and hypes. Trends Cancer. 2020;S2405–8033:30257–30250. transferases from searching genomic and proteomic sequences. Nucleic 4. Pilié PG, Tang C, Mills GB, Yap TA. State-of-the-art strategies for targeting Acids Res. 1999;27:3138–45. the DNA damage response in cancer. Nat Rev Clin Oncol. 2019;16:81–104. 30. Hua S, Quan Y, Zhan M, Liao H, Li Y, Lu L. miR-125b-5p inhibits cell prolif- 5. Mateo J, Lord CJ, Serra V, Tutt A, Balmaña J, Castroviejo-Bermejo M, Cruz C, eration, migration, and invasion in hepatocellular carcinoma via targeting Oaknin A, Kaye SB, de Bono JS. A decade of clinical development of PARP TXNRD1. Cancer Cell Int. 2019;19:203. inhibitors in perspective. Ann Oncol. 2019;30:1437–47. 31. Zhou Q, Zhou LQ, Li SH, Yuan YW, Liu L, Wang JL, Wu DZ, Wu Y, Xin 6. Okugawa Y, Grady WM, Goel A. Epigenetic alterations in colorectal cancer: L. Identifcation of subtype-specifc genes signature by WGCNA for emerging biomarkers. Gastroenterology. 2015;149:1204-1225.e1212. prognostic prediction in difuse type gastric cancer. Aging (Albany NY). 7. Acunzo M, Romano G, Wernicke D, Croce CM. MicroRNA and cancer–a 2020;12:17418–35. brief overview. Adv Biol Regul. 2015;57:1–9. 32. Szklarczyk D, Morris JH, Cook H, Kuhn M, Wyder S, Simonovic M, Santos 8. Shen Z, Lin L, Cao B, Zhou C, Hao W, Ye D. LZTS2 promoter hypermethyla- A, Doncheva NT, Roth A, Bork P, et al. The STRING database in 2017: tion: a potential biomarker for the diagnosis and prognosis of laryngeal quality-controlled protein-protein association networks, made broadly squamous cell carcinoma. World J Surg Oncol. 2018;16:42. accessible. Nucleic Acids Res. 2017;45:D362-d368. 9. Chan JJ, Tay Y. Noncoding RNA:RNA regulatory networks in cancer. Int J 33. Zhou H, Chen L, Lei Y, Li T, Li H, Cheng X. Integrated analysis of tumor Mol Sci. 2018;19:1310. mutation burden and immune infltrates in endometrial cancer. Curr 10. Anastas JN, Moon RT. WNT signalling pathways as therapeutic targets in Probl Cancer. 2021;45(2):100660. cancer. Nat Rev Cancer. 2013;13:11–26. 34. Du Z, Ma K, Sun X, Li A, Wang H, Zhang L, Lin F, Feng X, Song J. Methyla- 11. Vicente-Dueñas C, Hauer J, Cobaleda C, Borkhardt A, Sánchez-García I. tion of RASSF1A gene promoter and the correlation with DNMT1 expres- Epigenetic priming in cancer initiation. Trends Cancer. 2018;4:408–17. sion that may contribute to esophageal squamous cell carcinoma. World 12. Zhao H, Wang S, Song C, Zha Y, Li L. The prognostic value of MGMT pro- J Surg Oncol. 2015;13:141. moter status by pyrosequencing assay for glioblastoma patients’ survival: 35. Barabási AL, Gulbahce N, Loscalzo J. Network medicine: a network-based a meta-analysis. World J Surg Oncol. 2016;14:261. approach to human disease. Nat Rev Genet. 2011;12:56–68. 13. Shi H, Chai P, Jia R, Fan X. Novel insight into the regulatory roles of diverse 36. Panneerdoss S, Eedunuri VK, Yadav P, Timilsina S, Rajamanickam S, Viswa- RNA modifcations: re-defning the bridge between transcription and nadhapalli S, Abdelfattah N, Onyeagucha BC, Cui X, Lai Z, et al. Cross-talk translation. Mol Cancer. 2020;19:78. among writers, readers, and erasers of m(6)A regulates cancer growth 14. Liu RJ, Long T, Li J, Li H, Wang ED. Structural basis for substrate binding and progression. Sci Adv. 2018;4:eaar8263. and catalytic mechanism of a human RNA:m5C NSun6. 37. He Y, Yu X, Li J, Zhang Q, Zheng Q, Guo W. Role of m(5)C-related regula- Nucleic Acids Res. 2017;45:6684–97. tory genes in the diagnosis and prognosis of hepatocellular carcinoma. 15. Lei L, Jiang Z, Zhang G, Cheng Q, Lu H. MGMT promoter methylation and Am J Transl Res. 2020;12:912–22. 1p/19q co-deletion of surgically resected pulmonary carcinoid and large- 38. Abdel-Hafz HA, Horwitz KB. Role of epigenetic modifcations in luminal cell neuroendocrine carcinoma. World J Surg Oncol. 2018;16:110. breast cancer. Epigenomics. 2015;7:847–62. 16. Dixit R, Singh G, Pandey M, Basu S, Bhartiya SK, Singh KK, Shukla VK. 39. Costa-Pinheiro P, Montezuma D, Henrique R, Jerónimo C. Diagnos- Association of methylenetetrahydrafolate reductase gene polymorphism tic and prognostic epigenetic biomarkers in cancer. Epigenomics. (MTHFR) in patients with gallbladder cancer. J Gastrointest Cancer. 2015;7:1003–15. 2016;47:55–60. 40. Wang Y, Zhang R, Wu D, Lu Z, Sun W, Cai Y, Wang C, Jin J. Epigenetic 17. Wu J, Zhang Y, Li M. Identifcation of methylation markers and diferen- change in kidney tumor: downregulation of histone acetyltransferase tially expressed genes with prognostic value in breast cancer. J Comput MYST1 in human renal cell carcinoma. J Exp Clin Cancer Res. 2013;32:8. Biol. 2019;26:1394–408. 41. Cheng Y, He C, Wang M, Ma X, Mo F, Yang S, Han J, Wei X. Targeting epige- 18. García-Vílchez R, Sevilla A, Blanco S. Post-transcriptional regulation by netic regulators for cancer therapy: mechanisms and advances in clinical cytosine-5 methylation of RNA. Biochim Biophys Acta Gene Regul Mech. trials. Signal Transduct Target Ther. 2019;4:62. 2019;1862:240–52. 42. Wouters BJ, Delwel R. Epigenetics and approaches to targeted epigenetic 19. Frye M, Blanco S. Post-transcriptional modifcations in development and therapy in acute myeloid leukemia. Blood. 2016;127:42–52. stem cells. Development. 2016;143:3871–81. 43. Iniguez AB, Alexe G, Wang EJ, Roti G, Patel S, Chen L, Kitara S, Conway A, 20. Dong Z, Cui H. The emerging roles of RNA modifcations in glioblastoma. Robichaud AL, Stolte B, et al. Resistance to epigenetic-targeted therapy Cancers (Basel). 2020;12:736. engenders tumor cell vulnerabilities associated with enhancer remod- 21. Trixl L, Lusser A. The dynamic RNA modifcation 5-methylcytosine and its eling. Cancer Cell. 2018;34:922-938.e927. emerging role as an epitranscriptomic mark. Wiley Interdiscip Rev RNA. 44. Atrian F, Lelièvre SA. Mining the epigenetic landscape of tissue polarity in 2019;10:e1510. search of new targets for cancer therapy. Epigenomics. 2015;7:1313–25. 22. Catania J, Fairweather DS. DNA methylation and cellular ageing. Mutat 45. Hashizume R. Epigenetic targeted therapy for difuse intrinsic pontine Res. 1991;256:283–93. glioma. Neurol Med Chir (Tokyo). 2017;57:331–42. 23. Bohnsack KE, Höbartner C, Bohnsack MT. Eukaryotic 5-methylcytosine 46. Estibariz I, Overmann A, Ailloud F, Krebes J, Josenhans C, Suerbaum S. (m5C) RNA : mechanisms, cellular functions, and links The core genome m5C methyltransferase JHP1050 (M.Hpy99III) plays an to disease. Genes (Basel). 2019;10:102. important role in orchestrating gene expression in Helicobacter pylori. 24. Yang T, Low JJA, Woon ECY. A general strategy exploiting m5C duplex- Nucleic Acids Res. 2019;47:2336–48. remodelling efect for selective detection of RNA and DNA m5C methyl- 47. Saghafnia S, Mina M, Riggi N, Hanahan D, Ciriello G. Pan-cancer land- transferase activity in cells. Nucleic Acids Res. 2020;48:e5. scape of aberrant DNA methylation across human tumors. Cell Rep. 25. Yang X, Yang Y, Sun BF, Chen YS, Xu JW, Lai WY, Li A, Wang X, Bhattarai 2018;25:1066-1080.e1068. DP, Xiao W, et al. 5-methylcytosine promotes mRNA export - NSUN2 48. Zhang Q, He Y, Luo N, Patel SJ, Han Y, Gao R, Modak M, Carotta S, as the methyltransferase and ALYREF as an m(5)C reader. Cell Res. Haslinger C, Kind D, et al. Landscape and dynamics of single immune 2017;27:606–25. cells in hepatocellular carcinoma. Cell. 2019;179:829-845.e820. 26. Dou L, Li X, Ding H, Xu L, Xiang H. Prediction of m5C Modifcations in RNA 49. Tse JWT, Jenkins LJ, Chionh F, Mariadason JM. Aberrant DNA meth- sequences by combining multiple sequence features. Mol Ther Nucleic ylation in colorectal cancer: what should we target? Trends Cancer. Acids. 2020;21:332–42. 2017;3:698–712. 27. Gama-Sosa MA, Slagel VA, Trewyn RW, Oxenhandler R, Kuo KC, Gehrke 50. Chen L, Wang D. Identifcation of potential CpG sites for oral squamous CW, Ehrlich M. The 5-methylcytosine content of DNA from human cell carcinoma diagnosis via integrated analysis of DNA methylation and tumors. Nucleic Acids Res. 1983;11:6883–94. gene expression. World J Surg Oncol. 2021;19:16. 28. Wu H, Zhang Y. Mechanisms and functions of Tet protein-mediated 51. Yoshinobu Takahashi HN. Keishi Makino: Prognostic value of isocitrate 5-methylcytosine oxidation. Genes Dev. 2011;25:2436–52. dehydrogenase 1, O6-methylguanine-DNA methyltransferase promoter methylation, and 1p19q co-deletion in Japanese malignant glioma patients. World J Surg Oncol. 2013;11:284. He et al. World J Surg Onc (2021) 19:224 Page 12 of 12

52. Sibbritt T, Patel HR, Preiss T. Mapping and signifcance of the mRNA molecular targets for hepatocellular carcinoma. Cancer Cell Int. methylome. Wiley Interdiscip Rev RNA. 2013;4:397–422. 2019;19:279. 53. Popis MC, Blanco S, Frye M. Posttranscriptional methylation of transfer 60. Pleasance ED, Cheetham RK, Stephens PJ, McBride DJ, Humphray SJ, and ribosomal RNA in stress response pathways, cell diferentiation, and Greenman CD, Varela I, Lin ML, Ordóñez GR, Bignell GR, et al. A compre- cancer. Curr Opin Oncol. 2016;28:65–71. hensive catalogue of somatic mutations from a human cancer genome. 54. Lu L, Zhu G, Zeng H, Xu Q, Holzmann K. High tRNA transferase NSUN2 Nature. 2010;463:191–6. gene expression is associated with poor prognosis in head and neck 61. Krepischi AC, Pearson PL, Rosenberg C. Germline copy number variations squamous carcinoma. Cancer Invest. 2018;36:246–53. and cancer predisposition. Future Oncol. 2012;8:441–50. 55. Cui X, Liang Z, Shen L, Zhang Q, Bao S, Geng Y, Zhang B, Leo V, Vardy LA, 62. Zhao X, Liu X, Zhang A, Chen H, Huo Q, Li W, Ye R, Chen Z, Liang L, Liu Lu T, et al. 5-Methylcytosine RNA methylation in Arabidopsis Thaliana. Mol QA, et al. The correlation of copy number variations with longevity in Plant. 2017;10:1387–99. a genome-wide association study of Han Chinese. Aging (Albany NY). 56. Haag S, Warda AS, Kretschmer J, Günnigmann MA, Höbartner C, Bohnsack 2018;10:1206–22. MT. NSUN6 is a human RNA methyltransferase that catalyzes formation of 63. Zhou Y, Wang Y, Zhou W, Chen T, Wu Q, Chutturghoon VK, Lin B, Geng m5C72 in specifc tRNAs. RNA. 2015;21:1532–43. L, Yang Z, Zhou L, Zheng S. YAP promotes multi-drug resistance and 57. Van Haute L, Lee SY, McCann BJ, Powell CA, Bansal D, Vasiliauskaitė L, Gar- inhibits autophagy-related cell death in hepatocellular carcinoma via the one C, Shin S, Kim JS, Frye M, et al. NSUN2 introduces 5-methylcytosines RAC1-ROS-mTOR pathway. Cancer Cell Int. 2019;19:179. in mammalian mitochondrial tRNAs. Nucleic Acids Res. 2019;47:8720–33. 64. Wang JZ, Zhu W, Han J, Yang X, Zhou R, Lu HC, Yu H, Yuan WB, Li PC, Tao J, 58. Janin M, Ortiz-Barahona V, de Moura MC, Martínez-Cardús A, Llinàs-Arias et al. The role of the HIF-1alpha/ALYREF/PKM2 axis in glycolysis and tum- P, Soler M, Nachmani D, Pelletier J, Schumann U, Calleja-Cervantes ME, origenesis of bladder cancer. Cancer Commun (Lond). 2021;41(7):560–75. et al. Epigenetic loss of RNA-methyltransferase NSUN5 in glioma targets ribosomes to drive a stress adaptive translational program. Acta Neuro- pathol. 2019;138:1053–74. Publisher’s Note 59. Li P, Xu Y, Zhang Q, Li Y, Jia W, Wang X, Xie Z, Liu J, Zhao D, Shao M, et al. Springer Nature remains neutral with regard to jurisdictional claims in pub- Evaluating the role of RAD52 and its interactors as novel potential lished maps and institutional afliations.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions